Pync Collar — Thermal Defect Analysis and LTE Architecture Recommendation
Advisory memo from prospective CTO during technical diligence. For CEO, CPO, hardware team, and firmware team.
Confidential — Internal
Executive Summary
For decision: Drop LTE from v1 collar due to unresolved thermal safety defect.
Context: Current design heats collar shell above 40°C against pet skin in poor cellular signal. Proposed software mitigation addresses wrong root cause.
Key actions within 2 weeks: Disable LTE in pilot firmware. Notify pilot users. Commission independent thermal testing. Decide v1 cellular strategy.
Strategic implication: v2 with cellular is 9-15 months with appropriate silicon. v1 ships BLE-only and competes on health intelligence, not GPS tracking.
Summary
The current collar design exhibits a thermal defect: sustained LTE transmission in poor-signal conditions heats the collar shell above 40°C in contact with pets and humans. The proposed software mitigation (reactive throttling of IMU sampling rate when temperature exceeds thresholds) does not address the root cause and is insufficient for a safety-critical pet-contact product.
This memo recommends dropping LTE from v1 and revisiting cellular connectivity for v2 using appropriate silicon and data patterns. The recommendation is driven by three factors: (1) product positioning does not require LTE, (2) current implementation reproduces known anti-patterns in wearable thermal design, (3) personal safety risk to pets and liability exposure to the company are substantial and not adequately mitigated by the proposed software fix.
The Defect as Flagged
Per documentation from the hardware/firmware team:
- Collar has LTE radio for data upload when both phone (BLE) and dock (BLE) are unreachable
- Data collection rate: 10KB every 1 second, uploaded every 5 seconds
- In poor cellular signal, LTE radio boosts transmit power, drawing sustained current
- Collar shell reaches above 40°C in contact with pets and humans
- Proposed mitigation: firmware state machine that reduces IMU sampling rate (100Hz → 50Hz → 25Hz) and disables audio capture at 35°C and 37°C thresholds
Data Architecture Inconsistency with Product Positioning
Beyond the thermal issue, the data upload pattern indicates a separate architectural concern worth surfacing.
10KB every 1 second, uploaded every 5 seconds = 10 KB/sec sustained = 80 kbps continuous.
This is three to four orders of magnitude higher than comparable health-intelligence wearables:
- Apple Watch heart rate and activity: ~1-5 KB/minute
- Oura Ring continuous monitoring: ~10-50 KB/hour
- Whoop 4.0 all-day monitoring: ~100-500 KB/hour
- Fi Series 3 dog collar: GPS + activity summary, <10 KB/hour
The product positioning is on-device behavioral classification on Apollo3. Correctly implemented, on-device ML produces compact classification outputs and feature vectors, not continuous raw sensor telemetry. The current upload pattern suggests one of:
- Raw IMU and audio data is being uploaded, with cloud-side classification — defeating the Apollo3 architectural choice.
- On-device ML is not actually running, or not producing usable compact outputs.
- The measurement is wrong and needs auditing.
All three require investigation. The thermal issue and the data rate issue share a common cause: sustained LTE activity. Resolving the data architecture is prerequisite to any cellular strategy, v1 or v2.
Why the Proposed Mitigation Is Insufficient
Reactive, not preventive. The throttle activates after the device has already reached 35°C. Pet has been exposed to the rising temperature before the software response.
Does not address the root cause. LTE radio power consumption in poor signal is the heat source. Reducing IMU sampling rate does not reduce LTE activity. The radio continues transmitting at boosted power, continuing to heat the device.
No hard cutoff. The flowchart contains no absolute temperature threshold at which LTE is disabled. If the throttle state machine fails, if the thermal model is wrong, or if thermal response lags the actual heat source, there is no last-resort protection.
No user visibility. The device has no mechanism to alert the owner that the collar is in overheat mode. The owner does not know to check on their pet.
Unclear temperature measurement location. The thresholds (33/35/37°C) are not specified as MCU, battery, or shell temperature. Pet-contact surface temperature is not measured directly. MCU die temperature can differ from shell contact temperature by 5-10°C.
Tight hysteresis. 2°C hysteresis (33/35/37) will cause state oscillation in marginal thermal environments. Not a safety issue but indicates the thermal model is not well-characterized.
Thermal Safety Thresholds for Pet-Contact Devices
Sustained skin contact temperature thresholds:
| Temperature | Effect |
|---|---|
| <37°C | Safe, indistinguishable from body temperature |
| 37-40°C | Warm but generally safe for sustained contact |
| 40-43°C | Uncomfortable for humans; risk of low-temperature burns over hours |
| 43-45°C | Low-temperature burn range for mammalian skin over extended exposure |
| >45°C | Acute burn risk |
Pet-specific considerations that worsen the risk profile:
- Fur traps heat against skin, reducing convective cooling
- Pets cannot remove the collar when uncomfortable
- Continuous contact over weeks/months can cause chronic dermatitis, fur loss, pressure sores
- Veterinary literature documents burns from heating devices at sustained temperatures as low as 43-44°C
- Thin-skinned collar-contact areas are more vulnerable than other body regions
Current design allows 40°C+ shell temperature during normal operation. This is not adequate safety margin for a pet-contact device.
Ambient Temperature Context
Thailand ambient averages 28-32°C year-round. Collar operating at 40°C shell temperature in 30°C ambient is a smaller thermal delta than would be observed in lower-ambient conditions.
For US launch (20°C winter ambient common), sustained LTE activity in poor signal may drive shell temperatures meaningfully above 40°C — potentially 45°C+. Thailand pilot data is likely a lower bound on the thermal behavior, not a representative estimate. Thermal testing protocol must include low-ambient conditions, not only Thailand-representative conditions.
Comparison to Smartwatch Precedent
Smartwatches (Apple Watch, Garmin, Fitbit, Samsung) ship cellular wearables successfully. Industry-standard solutions exist for this class of device. The current implementation diverges from these patterns in several specific ways:
1. Silicon choice. Apollo3 Blue was chosen for ultra-low-power BLE. It has no integrated cellular modem, no vendor-validated thermal behavior for cellular operation, no silicon-level power management coordinating MCU and LTE tasks. Nordic nRF9160/9161 is the industry-standard choice for LTE-M wearables at this class — integrated Cortex-M33 + LTE-M/NB-IoT modem, designed for battery-powered cellular IoT. Used in Fitbark and numerous other wearables. Apollo3 with a bolted-on LTE module is not the right architecture for this use case.
2. Enclosure. Pet collars are plastic (cost, weight, injection molding). Smartwatches use metal chassis as heat spreaders. Pet collar plastic enclosure concentrates heat at the module rather than dissipating it.
3. Data pattern. Smartwatches transmit at idle with short bursts. Current Pync collar uploads 10KB every 5 seconds continuously. This is orders of magnitude more radio activity than smartwatch idle patterns. Continuous radio activity generates continuous heat.
4. Thermal safety margin. Apple Watch thermal spec allows case temperatures to ~43°C. Humans notice discomfort and remove the device. Pet collars have less thermal headroom, not more, because pets cannot remove the device and fur traps heat.
5. Failure mode (poor signal). Standard cellular wearable failure mode is LTE power boost in poor signal. Smartwatches address this by limiting maximum transmit power budget and backing off data rate. Current Pync design does neither; software fix does not address LTE power.
Architecture Mismatch Between Design Intent and Implementation
Contractor stack documentation (referenced in attachments) describes a BLE-first architecture consistent with Apollo3 Blue’s design purpose. The current always-on LTE implementation diverges from this original design intent. The thermal issue is a consequence of using ultra-low-power BLE silicon in a continuous-cellular data pattern it was not specified for — not a consequence of the original contractor design.
Product Positioning and LTE Necessity
Current product positioning has emphasized health and activity intelligence, not GPS tracking. This positioning does not require LTE.
Use cases by connectivity availability:
| Scenario | Phone BLE Available? | Dock BLE Available? | LTE Needed? |
|---|---|---|---|
| Pet with owner away from home | Yes | No | No |
| Pet at home with dock | Maybe | Yes | No |
| Pet at home, dock offline/WiFi down | Likely yes | No | No |
| Pet at daycare / walker / vet | No | No | Only for real-time sync |
| Pet lost / escaped | No | No | Yes, if marketing as GPS tracker |
LTE provides value only in the bottom two rows. For a health intelligence product, asynchronous sync upon reconnection is acceptable — the user sees their pet’s day when they reconnect, not in real-time. The lost-dog use case is the primary LTE justification, and it is not the positioning Pync has chosen.
Regulatory and Liability Considerations
United States
- FCC Part 15 (BLE, unintentional radiator), FCC Part 22/24/27 (LTE)
- IEC 62368-1 for product safety, including accessible surface temperature limits
- CPSC Section 15(b) reporting obligations for products with defects that could create substantial product hazards. Known thermal defect plus continued shipment triggers reporting duty.
- State product liability law: Strict liability standard in most states for design defects. Known defect aggravates damages at trial.
- Pet products are not FDA-regulated, but veterinary health claims trigger additional marketing restrictions.
European Union
- CE marking under Radio Equipment Directive 2014/53/EU and Low Voltage Directive, incorporating IEC 62368-1 thermal limits
- General Product Safety Regulation (GPSR), effective December 2024, creates reporting obligations for unsafe consumer products
- Member state consumer protection laws apply; no unified EU regulation for pet products
Thailand
- NBTC Type Approval for RF devices (BLE and LTE separately certified)
- Consumer Protection Act and Liability for Damages Arising from Unsafe Products Act B.E. 2551 (2008) apply to products sold in Thailand
- 2008 Act creates strict liability for manufacturers regardless of fault. Claimants recover damages without proving negligence.
Industry Reference Standards for Pet Wearables
No dedicated ISO standard exists for pet wearables. Reasonable reference standards:
- IEC 60601-1 medical device thermal limits (applied by analogy)
- ISO 10993 biocompatibility for skin-contact materials
- IEC 62368-1 Annex A thermal limits for accessible parts
Pilot Liability Exposure
If any of the 300 pilot dogs has experienced skin injury from the collar, liability exposure under Thailand’s 2008 Act is immediate and does not require proof of negligence. Veterinary records created during the pilot period are discoverable. Proceeding to scale production or US launch without resolving the thermal defect materially increases liability on future incidents because the defect is now known and documented.
Firmware IP and Remediation Capacity
Two additional items affecting the v2 path and company readiness for Series A diligence:
IP ownership audit. The current firmware was built by a low-cost Chinese contractor. Work-for-hire provisions in Chinese contractor agreements are often ambiguous. Pync’s ownership of the firmware IP should be confirmed by counsel before scaling, before Series A due diligence, and before any v2 rebuild that might reuse existing code.
Remediation capacity. If v2 architecture requires new silicon (Nordic nRF9160) and new firmware, the existing contractor likely cannot deliver consumer-product-grade work on that stack. Budget and timeline for new firmware partner or in-house hire must be included in v2 roadmap. Rough order of magnitude: $300-800K for a reputable US or EU firmware consultancy to deliver v2-grade firmware on Nordic SoC, 6-9 months.
Benefits of Dropping LTE in v1
Eliminates the thermal defect. The heat source is removed. No software workaround required. No hardware thermal redesign required for v1 ship.
BOM reduction. LTE module + eSIM + antenna + RF components: approximately 40-75K saved. At 50,000 units: $400-750K.
Eliminates cellular connectivity subscription cost. Consumer cellular IoT runs 60-300K/year eliminated. At 50,000 devices: 3M/year eliminated.
Improves battery life significantly. LTE radios are the largest battery drain in cellular wearables. Dropping LTE extends battery life by days.
Simpler regulatory surface. BLE-only devices require FCC/CE/NBTC radio certification only. LTE devices additionally require PTCRB certification and carrier approvals per market. Months and tens of thousands of dollars saved per market.
Smaller, lighter device. LTE module + larger battery (to support LTE bursts) adds weight and volume. Pet owners prefer smaller, lighter collars.
Simpler firmware and architecture. The collar speaks BLE only. No cellular stack, no LTE state machine, no thermal throttle complexity. Lower firmware defect surface.
Matches product positioning. Health intelligence for pets does not require always-on connectivity. Oura (human wearable, BLE + WiFi sync, no cellular) is a positioning precedent.
What Is Lost by Dropping LTE
Cannot market as GPS tracker or “find my lost dog.” Marketing must not imply this capability. If any current marketing implies GPS tracking, it must be updated.
Data gap when pet is away from phone and dock. Collar buffers locally, syncs on reconnection. For health insights, invisible to users. For real-time alerts (anxiety spikes, health events), a real limitation.
Cannot send remote commands to collar when owner is away. BLE range only for commands (beep collar, trigger sync).
Competitive differentiation. Fi Series 3 at 250 BLE-only must justify the price premium through better health insights, better battery life, better app, better ML. Marketing and product must deliver on this.
Recommendation
Ship v1 BLE-only. Revisit cellular connectivity for v2 with appropriate silicon and data patterns.
Specific actions:
- Halt plans to ship additional collars with LTE enabled until the thermal defect is addressed
- Audit the 300 pilot collars in Thailand: are pets currently being exposed to 40°C+ temperatures? If yes, immediate notification to pilot users is required
- Disable LTE in pilot firmware as an interim measure while hardware path is decided
- Confirm marketing does not imply GPS tracking or lost-dog recovery; adjust if it does
- Commission independent thermal testing on the current collar (Intertek, TÜV, UL, or specialized wearables lab) to quantify actual shell temperature under worst-case conditions — needed regardless of v1 LTE decision because baseline data is required
- Budget for v2 redesign to use Nordic nRF9160 or equivalent integrated LTE-M SoC if cellular is strategically important
Pilot User Notification Protocol
“Immediate notification to pilot users” requires specification to be meaningful:
- Plain-language disclosure of the thermal issue in Thai, not technical English
- Option to pause participation pending resolution
- Veterinary check-up offered and paid for by Pync for any pet showing signs of skin irritation or fur loss at the collar contact area
- Documentation of notification sent and responses received, for regulatory and legal record
- Firmware update disabling LTE deployed concurrent with notification
Roadmap for V2 Cellular (If Pursued)
If post-v1 customer research confirms demand for cellular features (lost-dog recovery, real-time alerts when pet is away), the path forward is:
Silicon: Nordic nRF9160 or nRF9161. Integrated Cortex-M33 + LTE-M/NB-IoT modem. Vendor-validated thermal behavior for battery-powered cellular IoT. Industry-standard for this device class.
Data pattern: Pre-aggregate on-device. Transmit summaries every 5-15 minutes, not raw data every 5 seconds. Reduce data volume by 50-100x. LTE radio active for seconds, not continuous.
Thermal design: Enclosure redesigned for cellular thermal dissipation. Heat spreader between LTE module and pet-contact surface. RF module placement away from skin contact. Thermal gasket or equivalent.
Safety layers:
- Firmware-level thermal watchdog independent of main application, disables LTE at absolute maximum temperature
- Hardware-level thermal fuse or PTC on battery and LTE power rail
- User-visible notification when collar enters thermal protection mode
- Absolute maximum shell temperature specification (e.g., 40°C contact surface) enforced by multiple independent mechanisms
Certification: Full thermal and safety testing at third-party lab before production. Pet wearables testing protocols (consult with veterinary engineering advisors).
Firmware capacity: New silicon and new architecture requires firmware partner selection. Existing contractor unlikely to be the right fit. Budget $300-800K and 6-9 months for reputable firmware consultancy.
Timeline: Realistically 9-15 months for a v2 collar with cellular done correctly, parallel to v1 ship and market validation.
Decision Request
The founders and hardware team should decide within the next 2 weeks:
- Accept the recommendation to drop LTE from v1, or articulate specific product requirements that justify keeping LTE in v1 despite the thermal risk
- If keeping LTE: commit to hardware redesign timeline and independent thermal validation before shipping additional units
- If dropping LTE: confirm marketing and product positioning alignment; update pilot firmware to disable LTE
- Either path: authorize independent thermal testing of current collar to establish baseline measurement
- Commission counsel review of firmware IP ownership from contractor agreements
Every week the 300 pilot collars continue operating with LTE enabled and without independent thermal testing increases the window for potential pet injury and the company’s liability exposure. The recommended interim measure (disable LTE in pilot firmware) can be implemented in days and has no cost. Deferring this specific action is not justified by any competing priority.
Attachments Referenced
- “Overheating situation description” document from firmware team (describes the defect)
- “The software working logic” flowchart (shows the proposed throttle state machine)
- Earlier stack documentation from contractor team (confirms Apollo3 Blue silicon, BLE-first original architecture intent)
This memo represents a prospective CTO’s technical assessment during diligence. Final product decisions are founder authority. The prospective CTO’s role at this stage is to ensure decisions are made with full understanding of technical, safety, and liability implications.